The long term objective of this proposal is the development of an accurate method of structure prediction for biologically-important, homologous proteins to elucidate their function and to aid in design of therapeutic drugs and of profitable mutations. This goal is supported by the careful testing of predictions against crystal structures and spectral data. The specific aims are to test the prediction of the membrane-active, hemolytic toxin structures against the crystal structures we are determining and to extend predictions and testing of the method to the larger, homologous systems of the kringles from plasminogen and the phycobiliproteins from blue-green alga. Health-related examples of the method are the structure prediction of plasminogen (a primary physiological fibrinolytic agent involved in the maintenance of blood fluidity) from the kringle and serine protease structures (this proposal), of human dihydrofolate reductase (a protein implicated in human cancer) from the E. coli crystal structure, of new immonoglobulins from known immunoglobulin structures, and of renin, important in blood pressure control (future research). The prediction method is based on computer graphics modelling of an unknown structure from a known homologous one coupled with global energy minimization with the program AMBER. Several pathways for minimization are chosen to sample the potential energy surface. The potential energy function has been tested against the well-determined (0.945 A resolution) structure of crambin. Unique to this proposal and critical to the development of the method is a correlation of the predicted structure with the crystal structure and with circular dichroism and Raman spectra, which are sensitive to secondary structure. Systems under study are models for protein-lipid interactions (hemolytic toxins), have well-characterized lysine-binding behavior (kringles) and have been crystallized in our laboratory. (We have developed a method for crystallizing the very soluble, basic toxins as well as kringle 4). Our intent is to refine the prediction method by progressing from small structures (5,000 MW) to larger and more complex structures (10,000 and 17,500 MW).